Label scraping for AutoFill.

Adds label scraping to AutoFill.  Infers labels from surrounding context of
input fields in the WebKit DOM.  Specific cases added are:
  - Text element immediately preceeding INPUT element.
    Eg. First name:<INPUT type="text" id="firstname" value="John"/>
  - Paragraph element containing text immediately preceeding INPUT element.
    Eg. <P>First name:</P><INPUT type="text" id="firstname" value="John"/>

BUG=33031
TEST=FormManagerTest.LabelsFromInferredParagraph,
     FormManagerTest.LabelsFromInferredText,
     FormStructureTest.HeuristicsLabelsOnly

Review URL: http://codereview.chromium.org/1380002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@42720 0039d316-1c4b-4281-b951-d872f2087c98

4 files changed, 277 insertions, 44 deletions

diff --git a/chrome/browser/autofill/form_structure_unittest.cc b/chrome/browser/autofill/form_structure_unittest.cc
index 1245229..a8283a5 100644
--- a/chrome/browser/autofill/form_structure_unittest.cc
+++ b/chrome/browser/autofill/form_structure_unittest.cc
@@ -384,4 +384,102 @@ TEST(FormStructureTest, HeuristicsSample6) {
   EXPECT_EQ(ADDRESS_HOME_ZIP, form_structure->field(5)->heuristic_type());
 }
 
+// Tests a sequence of FormFields where only labels are supplied to heuristics
+// for matching.  This works because FormField labels are matched in the case
+// that input element ids (or |name| fields) are missing.
+TEST(FormStructureTest, HeuristicsLabelsOnly) {
+  scoped_ptr<FormStructure> form_structure;
+  webkit_glue::FormFieldValues values;
+
+  values.method = ASCIIToUTF16("post");
+  values.elements.push_back(webkit_glue::FormField(ASCIIToUTF16("First Name"),
+                                                   string16(),
+                                                   string16(),
+                                                   ASCIIToUTF16("text"),
+                                                   WebInputElement::Text));
+  values.elements.push_back(webkit_glue::FormField(ASCIIToUTF16("Last Name"),
+                                                   string16(),
+                                                   string16(),
+                                                   ASCIIToUTF16("text"),
+                                                   WebInputElement::Text));
+  values.elements.push_back(webkit_glue::FormField(ASCIIToUTF16("EMail"),
+                                                   string16(),
+                                                   string16(),
+                                                   ASCIIToUTF16("text"),
+                                                   WebInputElement::Text));
+  values.elements.push_back(webkit_glue::FormField(ASCIIToUTF16("Phone"),
+                                                   string16(),
+                                                   string16(),
+                                                   ASCIIToUTF16("text"),
+                                                   WebInputElement::Text));
+  values.elements.push_back(webkit_glue::FormField(ASCIIToUTF16("Fax"),
+                                                   string16(),
+                                                   string16(),
+                                                   ASCIIToUTF16("text"),
+                                                   WebInputElement::Text));
+  values.elements.push_back(webkit_glue::FormField(ASCIIToUTF16("Address"),
+                                                   string16(),
+                                                   string16(),
+                                                   ASCIIToUTF16("text"),
+                                                   WebInputElement::Text));
+  values.elements.push_back(webkit_glue::FormField(ASCIIToUTF16("Address"),
+                                                   string16(),
+                                                   string16(),
+                                                   ASCIIToUTF16("text"),
+                                                   WebInputElement::Text));
+  values.elements.push_back(webkit_glue::FormField(ASCIIToUTF16("Zip code"),
+                                                   string16(),
+                                                   string16(),
+                                                   ASCIIToUTF16("text"),
+                                                   WebInputElement::Text));
+  values.elements.push_back(webkit_glue::FormField(string16(),
+                                                   ASCIIToUTF16("Submit"),
+                                                   string16(),
+                                                   ASCIIToUTF16("submit"),
+                                                   WebInputElement::Submit));
+  form_structure.reset(new FormStructure(values));
+  EXPECT_TRUE(form_structure->IsAutoFillable());
+
+  // Expect the correct number of fields.
+  ASSERT_EQ(8UL, form_structure->field_count());
+
+  // Check that heuristics are initialized as UNKNOWN_TYPE.
+  std::vector<AutoFillField*>::const_iterator iter;
+  size_t i;
+  for (iter = form_structure->begin(), i = 0;
+       iter != form_structure->end();
+       ++iter, ++i) {
+    // Expect last element to be NULL.
+    if (i == form_structure->field_count()) {
+      ASSERT_EQ(static_cast<AutoFillField*>(NULL), *iter);
+    } else {
+      ASSERT_NE(static_cast<AutoFillField*>(NULL), *iter);
+      EXPECT_EQ(UNKNOWN_TYPE, (*iter)->heuristic_type());
+    }
+  }
+
+  // Compute heuristic types.
+  form_structure->GetHeuristicAutoFillTypes();
+  ASSERT_EQ(8U, form_structure->field_count());
+
+  // Check that heuristics are no longer UNKNOWN_TYPE.
+  // First name.
+  EXPECT_EQ(NAME_FIRST, form_structure->field(0)->heuristic_type());
+  // Last name.
+  EXPECT_EQ(NAME_LAST, form_structure->field(1)->heuristic_type());
+  // Email.
+  EXPECT_EQ(EMAIL_ADDRESS, form_structure->field(2)->heuristic_type());
+  // Phone.
+  EXPECT_EQ(PHONE_HOME_WHOLE_NUMBER,
+      form_structure->field(3)->heuristic_type());
+  // Fax.  Note, we don't currently match fax.
+  EXPECT_EQ(UNKNOWN_TYPE, form_structure->field(4)->heuristic_type());
+  // Address.
+  EXPECT_EQ(ADDRESS_HOME_LINE1, form_structure->field(5)->heuristic_type());
+  // Address Line 2.
+  EXPECT_EQ(ADDRESS_HOME_LINE2, form_structure->field(6)->heuristic_type());
+  // Zip.
+  EXPECT_EQ(ADDRESS_HOME_ZIP, form_structure->field(7)->heuristic_type());
+}
+
 }  // namespace
diff --git a/chrome/renderer/form_manager.cc b/chrome/renderer/form_manager.cc
index 246c448..2752dcb 100644
--- a/chrome/renderer/form_manager.cc
+++ b/chrome/renderer/form_manager.cc
@@ -202,10 +202,10 @@ void FormManager::FormElementToFormData(WebFrame* frame,
     string16 form_control_type = input_element.formControlType();
     WebInputElement::InputType input_type = input_element.inputType();
     FormField field = FormField(label,
-                                                          name,
-                                                          value,
-                                                          form_control_type,
-                                                          input_type);
+                                name,
+                                value,
+                                form_control_type,
+                                input_type);
     form->fields.push_back(field);
   }
 }
@@ -221,5 +221,48 @@ string16 FormManager::LabelForElement(const WebInputElement& element) {
         return label.innerText();
     }
   }
-  return string16();
+
+  // Infer the label from context if not found in label element.
+  return FormManager::InferLabelForElement(element);
+}
+
+// static
+string16 FormManager::InferLabelForElement(const WebInputElement& element) {
+  string16 inferred_label;
+  WebNode previous = element.previousSibling();
+  if (!previous.isNull()) {
+    if (previous.isTextNode()) {
+      inferred_label = previous.nodeValue();
+      TrimWhitespace(inferred_label, TRIM_ALL, &inferred_label);
+    }
+
+    // If we didn't find text, check for previous paragraph.
+    // Eg. <p>Some Text</p><input ...>
+    // Note the lack of whitespace between <p> and <input> elements.
+    if (inferred_label.empty()) {
+      if (previous.isElementNode()) {
+        WebElement element = previous.toElement<WebElement>();
+        if (element.hasTagName("p")) {
+          inferred_label = element.innerText();
+          TrimWhitespace(inferred_label, TRIM_ALL, &inferred_label);
+        }
+      }
+    }
+
+    // If we didn't find paragraph, check for previous paragraph to this.
+    // Eg. <p>Some Text</p>   <input ...>
+    // Note the whitespace between <p> and <input> elements.
+    if (inferred_label.empty()) {
+      previous = previous.previousSibling();
+      if (!previous.isNull() && previous.isElementNode()) {
+        WebElement element = previous.toElement<WebElement>();
+        if (element.hasTagName("p")) {
+          inferred_label = element.innerText();
+          TrimWhitespace(inferred_label, TRIM_ALL, &inferred_label);
+        }
+      }
+    }
+  }
+
+  return inferred_label;
 }
diff --git a/chrome/renderer/form_manager.h b/chrome/renderer/form_manager.h
index a6805b2..4928838 100644
--- a/chrome/renderer/form_manager.h
+++ b/chrome/renderer/form_manager.h
@@ -80,6 +80,11 @@ class FormManager {
   // Returns the corresponding label for |element|.
   static string16 LabelForElement(const WebKit::WebInputElement& element);
 
+  // Infers corresponding label for |element| from surrounding context in the
+  // DOM.  Contents of preceeding <p> tag or preceeding text element found in
+  // the form.
+  static string16 InferLabelForElement(const WebKit::WebInputElement& element);
+
   // The map of form elements.
   WebFrameFormElementMap form_elements_map_;
 
diff --git a/chrome/renderer/form_manager_unittest.cc b/chrome/renderer/form_manager_unittest.cc
index 418afd8..c288d19 100644
--- a/chrome/renderer/form_manager_unittest.cc
+++ b/chrome/renderer/form_manager_unittest.cc
@@ -19,20 +19,19 @@ using WebKit::WebString;
 using webkit_glue::FormData;
 using webkit_glue::FormField;
 
-class FormManagerTest : public RenderViewTest {
- public:
-  FormManagerTest() {}
-};
+namespace {
+
+typedef RenderViewTest FormManagerTest;
 
 TEST_F(FormManagerTest, ExtractForms) {
   LoadHTML("<FORM name=\"TestForm\" action=\"http://cnn.com\" method=\"post\">"
-           "  <INPUT type=\"text\" id=\"firstname\" value=\"John\">"
-           "  <INPUT type=\"text\" id=\"lastname\" value=\"Smith\">"
-           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\">"
+           "  <INPUT type=\"text\" id=\"firstname\" value=\"John\"/>"
+           "  <INPUT type=\"text\" id=\"lastname\" value=\"Smith\"/>"
+           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\"/>"
            "</FORM>");
 
   WebFrame* web_frame = GetMainFrame();
-  ASSERT_TRUE(web_frame);
+  ASSERT_NE(static_cast<WebFrame*>(NULL), web_frame);
 
   FormManager form_manager;
   form_manager.ExtractForms(web_frame);
@@ -67,16 +66,16 @@ TEST_F(FormManagerTest, ExtractForms) {
 
 TEST_F(FormManagerTest, ExtractMultipleForms) {
   LoadHTML("<FORM name=\"TestForm\" action=\"http://cnn.com\" method=\"post\">"
-           "  <INPUT type=\"text\" id=\"firstname\" value=\"John\">"
-           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\">"
+           "  <INPUT type=\"text\" id=\"firstname\" value=\"John\"/>"
+           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\"/>"
            "</FORM>"
            "<FORM name=\"TestForm2\" action=\"http://zoo.com\" method=\"post\">"
-           "  <INPUT type=\"text\" id=\"lastname\" value=\"Smith\">"
-           "  <INPUT type=\"submit\" name=\"second\" value=\"Submit\">"
+           "  <INPUT type=\"text\" id=\"lastname\" value=\"Smith\"/>"
+           "  <INPUT type=\"submit\" name=\"second\" value=\"Submit\"/>"
            "</FORM>");
 
   WebFrame* web_frame = GetMainFrame();
-  ASSERT_TRUE(web_frame);
+  ASSERT_NE(static_cast<WebFrame*>(NULL), web_frame);
 
   FormManager form_manager;
   form_manager.ExtractForms(web_frame);
@@ -128,12 +127,12 @@ TEST_F(FormManagerTest, GetFormsAutocomplete) {
   // Form is not auto-completable due to autocomplete=off.
   LoadHTML("<FORM name=\"TestForm\" action=\"http://cnn.com\" method=\"post\""
            " autocomplete=off>"
-           "  <INPUT type=\"text\" id=\"firstname\" value=\"John\">"
-           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\">"
+           "  <INPUT type=\"text\" id=\"firstname\" value=\"John\"/>"
+           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\"/>"
            "</FORM>");
 
   WebFrame* web_frame = GetMainFrame();
-  ASSERT_TRUE(web_frame);
+  ASSERT_NE(static_cast<WebFrame*>(NULL), web_frame);
 
   FormManager form_manager;
   form_manager.ExtractForms(web_frame);
@@ -152,12 +151,12 @@ TEST_F(FormManagerTest, GetFormsAutocomplete) {
   LoadHTML("<FORM name=\"TestForm\" action=\"http://abc.com\" method=\"post\">"
            "  <INPUT type=\"text\" id=\"firstname\" value=\"John\""
            "   autocomplete=off>"
-           "  <INPUT type=\"text\" id=\"lastname\" value=\"Smith\">"
-           "  <INPUT type=\"submit\" name=\"reply\" value=\"Send\">"
+           "  <INPUT type=\"text\" id=\"lastname\" value=\"Smith\"/>"
+           "  <INPUT type=\"submit\" name=\"reply\" value=\"Send\"/>"
            "</FORM>");
 
   web_frame = GetMainFrame();
-  ASSERT_TRUE(web_frame);
+  ASSERT_NE(static_cast<WebFrame*>(NULL), web_frame);
 
   form_manager.Reset();
   form_manager.ExtractForms(web_frame);
@@ -188,13 +187,13 @@ TEST_F(FormManagerTest, GetFormsAutocomplete) {
 TEST_F(FormManagerTest, GetFormsElementsEnabled) {
   // The firstname element is not enabled due to disabled being set.
   LoadHTML("<FORM name=\"TestForm\" action=\"http://xyz.com\" method=\"post\">"
-           "  <INPUT disabled type=\"text\" id=\"firstname\" value=\"John\">"
-           "  <INPUT type=\"text\" id=\"lastname\" value=\"Smith\">"
-           "  <INPUT type=\"submit\" name=\"submit\" value=\"Send\">"
+           "  <INPUT disabled type=\"text\" id=\"firstname\" value=\"John\"/>"
+           "  <INPUT type=\"text\" id=\"lastname\" value=\"Smith\"/>"
+           "  <INPUT type=\"submit\" name=\"submit\" value=\"Send\"/>"
            "</FORM>");
 
   WebFrame* web_frame = GetMainFrame();
-  ASSERT_TRUE(web_frame);
+  ASSERT_NE(static_cast<WebFrame*>(NULL), web_frame);
 
   FormManager form_manager;
   form_manager.ExtractForms(web_frame);
@@ -224,13 +223,13 @@ TEST_F(FormManagerTest, GetFormsElementsEnabled) {
 
 TEST_F(FormManagerTest, FindForm) {
   LoadHTML("<FORM name=\"TestForm\" action=\"http://buh.com\" method=\"post\">"
-           "  <INPUT type=\"text\" id=\"firstname\" value=\"John\">"
-           "  <INPUT type=\"text\" id=\"lastname\" value=\"Smith\">"
-           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\">"
+           "  <INPUT type=\"text\" id=\"firstname\" value=\"John\"/>"
+           "  <INPUT type=\"text\" id=\"lastname\" value=\"Smith\"/>"
+           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\"/>"
            "</FORM>");
 
   WebFrame* web_frame = GetMainFrame();
-  ASSERT_TRUE(web_frame);
+  ASSERT_NE(static_cast<WebFrame*>(NULL), web_frame);
 
   FormManager form_manager;
   form_manager.ExtractForms(web_frame);
@@ -273,13 +272,13 @@ TEST_F(FormManagerTest, FindForm) {
 
 TEST_F(FormManagerTest, FillForm) {
   LoadHTML("<FORM name=\"TestForm\" action=\"http://buh.com\" method=\"post\">"
-           "  <INPUT type=\"text\" id=\"firstname\">"
-           "  <INPUT type=\"text\" id=\"lastname\">"
-           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\">"
+           "  <INPUT type=\"text\" id=\"firstname\"/>"
+           "  <INPUT type=\"text\" id=\"lastname\"/>"
+           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\"/>"
            "</FORM>");
 
   WebFrame* web_frame = GetMainFrame();
-  ASSERT_TRUE(web_frame);
+  ASSERT_NE(static_cast<WebFrame*>(NULL), web_frame);
 
   FormManager form_manager;
   form_manager.ExtractForms(web_frame);
@@ -352,13 +351,13 @@ TEST_F(FormManagerTest, FillForm) {
 
 TEST_F(FormManagerTest, Reset) {
   LoadHTML("<FORM name=\"TestForm\" action=\"http://cnn.com\" method=\"post\">"
-           "  <INPUT type=\"text\" id=\"firstname\" value=\"John\">"
-           "  <INPUT type=\"text\" id=\"lastname\" value=\"Smith\">"
-           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\">"
+           "  <INPUT type=\"text\" id=\"firstname\" value=\"John\"/>"
+           "  <INPUT type=\"text\" id=\"lastname\" value=\"Smith\"/>"
+           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\"/>"
            "</FORM>");
 
   WebFrame* web_frame = GetMainFrame();
-  ASSERT_TRUE(web_frame);
+  ASSERT_NE(static_cast<WebFrame*>(NULL), web_frame);
 
   FormManager form_manager;
   form_manager.ExtractForms(web_frame);
@@ -378,14 +377,57 @@ TEST_F(FormManagerTest, Reset) {
 TEST_F(FormManagerTest, Labels) {
   LoadHTML("<FORM name=\"TestForm\" action=\"http://cnn.com\" method=\"post\">"
            "  <LABEL for=\"firstname\"> First name: </LABEL>"
-           "    <INPUT type=\"text\" id=\"firstname\" value=\"John\">"
+           "    <INPUT type=\"text\" id=\"firstname\" value=\"John\"/>"
            "  <LABEL for=\"lastname\"> Last name: </LABEL>"
-           "    <INPUT type=\"text\" id=\"lastname\" value=\"Smith\">"
-           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\">"
+           "    <INPUT type=\"text\" id=\"lastname\" value=\"Smith\"/>"
+           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\"/>"
+           "</FORM>");
+
+  WebFrame* web_frame = GetMainFrame();
+  ASSERT_NE(static_cast<WebFrame*>(NULL), web_frame);
+
+  FormManager form_manager;
+  form_manager.ExtractForms(web_frame);
+
+  std::vector<FormData> forms;
+  form_manager.GetForms(&forms, FormManager::REQUIRE_NONE);
+  ASSERT_EQ(1U, forms.size());
+
+  const FormData& form = forms[0];
+  EXPECT_EQ(ASCIIToUTF16("TestForm"), form.name);
+  EXPECT_EQ(GURL(web_frame->url()), form.origin);
+  EXPECT_EQ(GURL("http://cnn.com"), form.action);
+
+  const std::vector<FormField>& fields = form.fields;
+  ASSERT_EQ(3U, fields.size());
+  EXPECT_EQ(FormField(ASCIIToUTF16("First name:"),
+                      ASCIIToUTF16("firstname"),
+                      ASCIIToUTF16("John"),
+                      ASCIIToUTF16("text"),
+                      WebInputElement::Text), fields[0]);
+  EXPECT_EQ(FormField(ASCIIToUTF16("Last name:"),
+                      ASCIIToUTF16("lastname"),
+                      ASCIIToUTF16("Smith"),
+                      ASCIIToUTF16("text"),
+                      WebInputElement::Text), fields[1]);
+  EXPECT_EQ(FormField(string16(),
+                      ASCIIToUTF16("reply-send"),
+                      ASCIIToUTF16("Send"),
+                      ASCIIToUTF16("submit"),
+                      WebInputElement::Submit), fields[2]);
+}
+
+TEST_F(FormManagerTest, LabelsFromInferredText) {
+  LoadHTML("<FORM name=\"TestForm\" action=\"http://cnn.com\" method=\"post\">"
+           "  First name:"
+           "    <INPUT type=\"text\" id=\"firstname\" value=\"John\"/>"
+           "  Last name:"
+           "    <INPUT type=\"text\" id=\"lastname\" value=\"Smith\"/>"
+           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\"/>"
            "</FORM>");
 
   WebFrame* web_frame = GetMainFrame();
-  ASSERT_TRUE(web_frame);
+  ASSERT_NE(static_cast<WebFrame*>(NULL), web_frame);
 
   FormManager form_manager;
   form_manager.ExtractForms(web_frame);
@@ -417,3 +459,48 @@ TEST_F(FormManagerTest, Labels) {
                       ASCIIToUTF16("submit"),
                       WebInputElement::Submit), fields[2]);
 }
+
+TEST_F(FormManagerTest, LabelsFromInferredParagraph) {
+  LoadHTML("<FORM name=\"TestForm\" action=\"http://cnn.com\" method=\"post\">"
+           "  <P>First name:</P><INPUT type=\"text\" "
+           "                           id=\"firstname\" value=\"John\"/>"
+           "  <P>Last name:</P>"
+           "    <INPUT type=\"text\" id=\"lastname\" value=\"Smith\"/>"
+           "  <INPUT type=\"submit\" name=\"reply-send\" value=\"Send\"/>"
+           "</FORM>");
+
+  WebFrame* web_frame = GetMainFrame();
+  ASSERT_NE(static_cast<WebFrame*>(NULL), web_frame);
+
+  FormManager form_manager;
+  form_manager.ExtractForms(web_frame);
+
+  std::vector<FormData> forms;
+  form_manager.GetForms(&forms, FormManager::REQUIRE_NONE);
+  ASSERT_EQ(1U, forms.size());
+
+  const FormData& form = forms[0];
+  EXPECT_EQ(ASCIIToUTF16("TestForm"), form.name);
+  EXPECT_EQ(GURL(web_frame->url()), form.origin);
+  EXPECT_EQ(GURL("http://cnn.com"), form.action);
+
+  const std::vector<FormField>& fields = form.fields;
+  ASSERT_EQ(3U, fields.size());
+  EXPECT_EQ(FormField(ASCIIToUTF16("First name:"),
+                      ASCIIToUTF16("firstname"),
+                      ASCIIToUTF16("John"),
+                      ASCIIToUTF16("text"),
+                      WebInputElement::Text), fields[0]);
+  EXPECT_EQ(FormField(ASCIIToUTF16("Last name:"),
+                      ASCIIToUTF16("lastname"),
+                      ASCIIToUTF16("Smith"),
+                      ASCIIToUTF16("text"),
+                      WebInputElement::Text), fields[1]);
+  EXPECT_EQ(FormField(string16(),
+                      ASCIIToUTF16("reply-send"),
+                      ASCIIToUTF16("Send"),
+                      ASCIIToUTF16("submit"),
+                      WebInputElement::Submit), fields[2]);
+}
+
+}  // namespace